Narcissus Search


Narcissus is an algorithm for ranking search results which deprioritises those results which are deemed "too popular". It is an artistic response to the prevalence of "positive feedback" mechanisms on the web, which prioritise popularity: hence Google finds pages which you already know, Twitter recommends that you follow existing celebrities, and Facebook invites you to like exactly what your friends already like.

In contrast, Narcissus helps you to find the "shadows" - the surprising, quirky and rejected - by deliberately hiding results that you or your peers are showing too much interest in.

A Video Demo

Alpha Testing

In the alpha test site:

Click here for Narcissus alpha testing site.

How the search engine works

You can see an example of Narcissus in the video below, or try it by following this link.

Narcissus can be applied to different kinds of data using different kinds of feedback, but in this example, we are searching stories within a textual database and we assess the popularity by measuring how many times searchers click on the results.

When users begin to search, the results are ordered according to whatever criteria the underlying search engine uses. Once a searcher clicks a result, the action triggers deprioritisation. The result will be relegated to appear at, or close to, the end of the result-list.

If a second searcher also insists on clicking the same result, Narcissus hides it altogether.

The result then spends a certain amount of time hidden until the algorithm allows it to appear in searches again.


The Narcissus algorithm was originally submitted to and won N.E.W.S. "Shadow Search Platform" competition.

Shadow Spot